I am learning the MPLS, here is the thing which is continusely confusing me. What I read is that the MPLS is far far better technique than the traditional IP routing because you dont have to look for the IP address and search in routing table and then the send the frame in the specified direction. Now here is the scenerio that confuses me a lot.
Lets suppose we are dealing with the MPLS domain.
Provider Edge Router Receives the packet without Label, and on the basis of destination address it will first assign the Label on that and forward it to the suitable interface, the other router will look for the label and then send it to the suitable interface. Now what is going on here packet received and then labels are searched in the suitable place lets suppose the label table and then assign the specific label on the packet and forward it to the specific interface.
Routing Table and IP scenerio
The packet comes and look for the destination IP address, if matched and then forward it to the interface assigned to that destination IP even no need to swap the IP address and assigned.
Now my question is that how the MPLS is technique is faster way of delivering the packets as the processor needs to do extra amount of work swapping and searching for suitable label (whcih is actually the same as for the routing table, search for the destination IP address) and extra work to remove or add the label??
a valid point indeed. Strictly speaking labels reside one level down compared to IP, thus insignificantly improving lookup speed. Nowadays both LFIB and FIB are being programmed to the hw, so the bare difference is a wee if not zero.
One of the scenarios where MPLS excel traditional lookups is when you've got multiple prefixes(FEC in MPLS lingo) covered by one single label. This, for example, gives us MPLS VPNs, BGP-free core.
There are other benefits like L2VPNs and Traffic Engineering.
Keep up your study and don't hesitate to share your thoughts, those are very welcome!